Abstract

In multi-agent systems, the individual agent must form coalitions to accomplish complex tasks. However, the centralized management model is not flexible in a dynamic environment. To overcome the restriction caused by central control, the paper presents a self-organizing dynamic coalition mechanism based on game theory. Firstly, we adopt a distributed network to communicate among agents, allowing agents to solve the real-time task assignment problem autonomously. Next, a non-cooperative game negotiation model is introduced to find the optimal strategy for each agent. Finally, the effectiveness of our mechanism is validated by comparing it with the traditional command model in three distributed networks. Experimental results indicate the proposed mechanism is capable to improve the system utility.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call